prohibit_parallel_limit_subselect_v2.rel10.patch
application/octet-stream
Filename: prohibit_parallel_limit_subselect_v2.rel10.patch
Type: application/octet-stream
Part: 0
Patch
Same data as JSON:
GET /api/v1/attachments/:id/patch
the parsed metadata as JSON — format, series position, per-file stats; never the diff bytes.
API reference →
Format: unified
Series: patch v2
| File | + | − |
|---|---|---|
| src/backend/optimizer/path/allpaths.c | 10 | 0 |
| src/backend/optimizer/plan/planner.c | 1 | 2 |
| src/include/optimizer/planner.h | 2 | 0 |
| src/test/regress/expected/select_parallel.out | 19 | 0 |
| src/test/regress/sql/select_parallel.sql | 5 | 0 |
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index fff2f4c..e122a52 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -612,7 +612,17 @@ set_rel_consider_parallel(PlannerInfo *root, RelOptInfo *rel,
* the SubqueryScanPath as not parallel-safe. (Note that
* set_subquery_pathlist() might push some of these quals down
* into the subquery itself, but that doesn't change anything.)
+ *
+ * We can't push sub-select containing LIMIT/OFFSET to workers as
+ * that can change the results depending on what is beneath LIMIT
+ * node. So, consider that case as parallel-unsafe.
*/
+ {
+ Query *subquery = castNode(Query, rte->subquery);
+
+ if (limit_needed(subquery))
+ return;
+ }
break;
case RTE_JOIN:
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index 1946f9e..a8a025c 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -121,7 +121,6 @@ static void preprocess_rowmarks(PlannerInfo *root);
static double preprocess_limit(PlannerInfo *root,
double tuple_fraction,
int64 *offset_est, int64 *count_est);
-static bool limit_needed(Query *parse);
static void remove_useless_groupby_columns(PlannerInfo *root);
static List *preprocess_groupclause(PlannerInfo *root, List *force);
static List *extract_rollup_sets(List *groupingSets);
@@ -2697,7 +2696,7 @@ preprocess_limit(PlannerInfo *root, double tuple_fraction,
* a key distinction: here we need hard constants in OFFSET/LIMIT, whereas
* in preprocess_limit it's good enough to consider estimated values.
*/
-static bool
+bool
limit_needed(Query *parse)
{
Node *node;
diff --git a/src/include/optimizer/planner.h b/src/include/optimizer/planner.h
index 2a4cf71..acb709a 100644
--- a/src/include/optimizer/planner.h
+++ b/src/include/optimizer/planner.h
@@ -46,6 +46,8 @@ extern bool is_dummy_plan(Plan *plan);
extern RowMarkType select_rowmark_type(RangeTblEntry *rte,
LockClauseStrength strength);
+extern bool limit_needed(Query *parse);
+
extern void mark_partial_aggref(Aggref *agg, AggSplit aggsplit);
extern Path *get_cheapest_fractional_path(RelOptInfo *rel,
diff --git a/src/test/regress/expected/select_parallel.out b/src/test/regress/expected/select_parallel.out
index 562eb2c..14420f4 100644
--- a/src/test/regress/expected/select_parallel.out
+++ b/src/test/regress/expected/select_parallel.out
@@ -573,6 +573,25 @@ explain (costs off)
Index Cond: (unique1 = 1)
(5 rows)
+-- LIMIT/OFFSET within sub-selects can't be pushed to workers.
+explain (costs off)
+select * from tenk1 where two in
+ (select two from tenk1 where stringu1 like '%AAAA' limit 3);
+ QUERY PLAN
+---------------------------------------------------------------
+ Hash Semi Join
+ Hash Cond: (tenk1.two = tenk1_1.two)
+ -> Gather
+ Workers Planned: 4
+ -> Parallel Seq Scan on tenk1
+ -> Hash
+ -> Limit
+ -> Gather
+ Workers Planned: 4
+ -> Parallel Seq Scan on tenk1 tenk1_1
+ Filter: (stringu1 ~~ '%AAAA'::text)
+(11 rows)
+
-- to increase the parallel query test coverage
EXPLAIN (analyze, timing off, summary off, costs off) SELECT * FROM tenk1;
QUERY PLAN
diff --git a/src/test/regress/sql/select_parallel.sql b/src/test/regress/sql/select_parallel.sql
index 1d29e96..c0bfc2d 100644
--- a/src/test/regress/sql/select_parallel.sql
+++ b/src/test/regress/sql/select_parallel.sql
@@ -204,6 +204,11 @@ set force_parallel_mode=1;
explain (costs off)
select stringu1::int2 from tenk1 where unique1 = 1;
+-- LIMIT/OFFSET within sub-selects can't be pushed to workers.
+explain (costs off)
+select * from tenk1 where two in
+ (select two from tenk1 where stringu1 like '%AAAA' limit 3);
+
-- to increase the parallel query test coverage
EXPLAIN (analyze, timing off, summary off, costs off) SELECT * FROM tenk1;